The extremely large numbers of conformations, accessible to peptides of size 10 to 20 residues, make them difficult to consider both theoretically and experimentally. Complete sets of their conformations are impossibly large to generate. A new approach to combine calculations with specific constraints has been developed. The constraining information could originate either from experiment or from a molecular model with specified interactions. This approach for considering random peptide conformations with constraints has been formulated in the context of helix-coil theory. Protein surfaces are being examined in order to develop methods to specify likely binding sites for peptides. Lattice fits of the target protein are extended to define an exterior shell. Each of the shell points is probed to evaluate the hydrophobicity of the protein residues in its neighborhood. The binding sites in 23 co-crystals of peptides bound to proteins are correctly identified by this procedure.